Transcript presentation on SPECT dynamic imaging

DYNAMIC IMAGE
RECONSTRUCTION
IN NUCLEAR MEDICINE
Ryan
O’Flaherty
Kyle Fontaine
Krystal Kerney
ACQUISITION TECHNIQUES
 Static
 Data acquisition starts after the radiotracer is distributed and settled
in the targeted tissues.
 Static SPECT provides one static 3D image of the distribution of the
radiotracer.
 Dynamic
 Data acquisition starts immediately after the injection of radiotracer.
 Dynamic SPECT provides a series of 3D images. Each image
represents the distribution of the radiotracer at a certain time.
 Dynamic images convey information about tracer movement through
different body tissues.
DATA ACQUISITION
 Input
 3D volume (body injected by the radiotracer).
—
P rocess
 Recording the activity of tracer in the 3D volume.
 Output
 A set of 2D projections taken from different angles (Sinogram).
IMAGE RECONSTRUCTION
 Input
 A set of 2D projections taken from different angles (Sinogram).
 Process
 Reconstructing the 3D volume back from the recorded projections.
 Output
 3D volume (radiotracer activity in the body tissues).
 Time Activity Curves (TACs) can be extracted from the
reconstructed time -dependent images for tissues in interest.
 Example of Myocardium Time Activity Curves
BSPLINES
 Instead of reconstructing a time independent-volume, we try
to estimate the input functions of tissues in interest .
 Time basis functions (B-splines) represent the temporal
behavior of radioactive tracer in the imaged tissues.
TASK
 Our task
 Generate desired time activity curves based off the time basis
functions (Bsplines) that we create.
 Generate Bsplines  Create TAC’s  Cluster TAC’s
 Example TAC to mimic:
MATHEMATICAL FORMULATION
 Similarly to static image reconstruction, a form of the
equation below is used:
 However, the case of dynamic imaging requires that the
imagined volume V_k be split up into several imagined
volumes, each representing the volume at a given time, m.
 C is an algorithmically determined coef ficient and f
represents the time dependent operator, the B -spline
MATHEMATICAL FORMULATION
 Thus, the new equation for the sinogram which will be used to
reconstruct V_k is below:






Where:
P is the sinogram vector
S is the system matrix
n is the number of pixels
m is the number of projections (and thus represents time)
K is the number of voxels
COEFFICIENT CALCULATION
* From Mamoud’s Presentation
-
This objective function x^2 is to be minimized based on the parameters C and f. This
allows for the verification of the coefficients
RESULTS
 An example of our time activity curves before clustering
 These were extracted from our reconstructed volumes (unable
to view)
Relative Activity (Arbitrary Units)
 If we could, it would be 72 volumes concatenated with each other
Time (s)
RESULTS
Relative Activity (Arbitrary Units)
 Shown below - the Bsplines and their corresponding TAC’s
Time (s)
Relative Activity (Arbitrary Units)
RESULTS
Time (s)
Relative Activity (Arbitrary Units)
RESULTS
Time (s)
Relative Activity (Arbitrary Units)
RESULTS
Time (s)
Relative Activity (Arbitrary Units)
NEW RESULTS
Time (s)
Relative Activity (Arbitrary Units)
NEW RESULTS
Time (s)
Relative Activity (Arbitrary Units)
NEW RESULTS
Time (s)
Relative Activity (Arbitrary Units)
NEW RESULTS
Time (s)
Relative Activity (Arbitrary Units)
BEST RESULT
Our result
Goal
Time (s)
REFINING RESULTS
 Take output clustered TAC  Feed back into C code as our
spline input
 Attempt to initialize the algorithm using different initial conditions
 Hope to find new, more accurate, local minimum
 Won’t necessarily ‘refine’ (can even have opposite effect)
Relative Activity (Arbitrary Units)
Relative Activity (Arbitrary Units)
REFINED RESULTS
Previous Best Result
Refined
Time (s)
Time (s)
Goal
Time (s)
Relative Activity (Arbitrary Units)
Relative Activity (Arbitrary Units)
REFINED RESULTS
Previous Result
Refined Result
Time (s)
Goal
Time (s)
Relative Activity (Arbitrary Units)
Relative Activity (Arbitrary Units)
REFINED RESULTS
Previous Result
Refined Result
Goal
Time (s)
SUMMARY
 Our task was to understand the concepts behind dynamic
imaging, and then reproduce given results.
 The team was able to recreate the given data using event ques
from a supplied sinogram. These events are translated into
knots, which represent the beginning, end, and inflection points
in a B-spline curve.
 Each of these curves were used as operators to affect
algorithmically determined constant coefficients. These
constant/cur ve pairs are summed to represent Time Activity
Curves, which denote the temporal dependence of specific
radiotracers within known tissues of a patient.
 Attempted to initialize the algorithm with different initial
conditions (clustered TAC outputs) in hopes for more accurate
minimization.
SUMMARY
 Our success was based on varying the B -splines we input, and
noting the influence of the changes on the clustered TAC’s.
 Notes
 Open ended B-spline necessary to avoid symmetrical TAC’s
 Such tendencies don’t make sense when considering the temporal activity
of radiotracer in tissue
 Clustering B-splines in either direction (more B-splines in the
beginning, or more B-splines later) caused irregular TAC’s
 Reducing the number of B-splines gave us TAC’s that more closely
mimicked our desired output
 However, increasing the number of B-splines didn’t necessarily cause
irregular TAC outputs.
 Result of refining
 Small but noticeable refinements for all 3 attempts
 More experimentation required (very time consuming)
REVIEW OF JOURNAL ARTICLE
(PART 1)
 Article title: The role of nuclear imaging in the failing heart:
myocardial blood flow, sympathetic innervation, and future
applications
 Journal: Heart Failure Reviews
 Year: 2010
NUCLEAR IMAGING IN THE FAILING
HEART
 Heart failure af fects approximately 5 million patients in the
United States!
SPECT/PET IMAGING IN THE FAILING
HEART
 Nuclear imaging is the only modality with suf ficient sensitivity
to assess blood flow and innervation of the failing heart.
 Innervation is excitation of the heart by nerve cells.
 SPECT is most commonly used for evaulation of myocardial
perfusion (blood flow to the heart).
 PET allows for quantification of myocardial blood flow.
 Both can be used for evaluation of diagnosis, treatment
options, and prognosis in heart failure patients.
SYMPATHETIC INNERVATION OF THE
HEART
 Sympathetic innervation (excitation) represents another
important parameter in patients with heart failure.
 Sympathetic nerve imaging with 123 -iodine.
metaiodobenzylguanidine (123-I MIBG) is often used for
assessment of cardiac innervation.
 Abnormal innervation is associated with increased mortality
and morbidity rates in patients with heart failure.
 123-I MIBG can be used to categorize patients by risk for
ventricular arrhythmias or sudden cardiac death.
POTENTIAL OF NUCLEAR IMAGING IN
HEART FAILURE
 Detailed information on several biological processes in heart
failure
 Myocardial blood flow
 Sympathetic innervation of the myocardium
 Myocardial perfusion imaging represents the mainstay of
cardiovascular radionuclide applications
 Sympathetic innervation imaging is increasingly used in
patients with heart failure
MYOCARDIAL BLOOD FLOW IN THE
FAILING HEART
 SPECT
 Well-established and safe imaging modality for the evaluation of
location, extent and severity of myocardial perfusion defects.
 3 commercially available SPECT tracers: 201Thallium, 99mTc tetrofosmin, and 99mTc-sestamibi.
 201Thallium
 99mTc-tetrofosmin
 99mTc-sestamibi
 PET
 Several PET tracers currently available for assessment of myocardial
perfusion
 2 approved for clinical use by the FDA
 N-13 Ammonia (13NH3)
 Rubidium-82 (82Rb)
 Both can be used for absolute quantification of myocardial blood flow
TABLE OF SPECT/PET TRACERS
DYNAMIC IMAGING AND TRACER
KINETICS
 Dynamic imaging with multiple time
frames requires a high count density
and advanced data processing
 For tracer kinetic analysis, arterial
input function and myocardial
kinetics are measured from regions
of interest in dynamic images
 Absolute flow quantification is
achieved by employing
compartmental modeling analysis to
the obtained time-activity curves.
 Various tracer kinetic models have
been established according to the
nature of each PET tracer
REVIEW OF JOURNAL ARTICLE
(PART 2)
Last time:
Introduction
Potential of nuclear imaging in heart failure
Myocardial blood flow in the failing heart.
Dynamic images and time-activity-curves
Today:
Sympathetic innervation in the failing heart
 Using SPECT
 Using PET
SYMPATHETIC INNERVATION IN THE
FAILING HEART
 The sympathetic nervous
system can do two things:
  strength of contraction
  amount of blood returned
 Sympathetic input:
speeds SA depolarization (HR
decreases AV delay (HR)
increases contractility in
contractile cells (SV)
In sum: sympathetic input increases
heart rate and degree of contraction
per beat
REGULATION OF THE SYMPATHETIC
NERVOUS SYSTEM
IMAGING WITH SPECT/PLANAR
 Radionuclide imaging of the norepinepherine analog
metaiodobenzylguanidine (MIGB) baleled with 123iodine (123-I).
 Planar and SPECT imaging are performed in the early
and late phase of the 123-I MIBG protocol.
 Planar images from Left-Anterior oblique view and provide
information on global sympathetic innervation pattern.
 SPECT images are used to assess regional abnormalities
in cardiac sympathetic innervation.
IMAGING WITH SPECT/PLANAR
IMAGING WITH PET
 In contrast to SPECT/Planar imaging, PET maps the
sympathetic nervous system with superior temporal
and spatial resolution.
 Spatial resolution of 4-7mm.
 Temporal resolution allows for development of dynamic
images which can be used to assess tracer kinetics.
 Can be used to quantify the absolute amount of tracer and
its time-dependent kinetics.
TWO CATEGORIES OF PET TRACERS
 Radiolabeled catecholamines
 Molecularly identical to endogenous neurotransmitters
 Undergo similar uptake, release and metabolic pathways
 Radiolabeled catecholamine analogs
 False neurotransmitters
 Follow the same uptake and release mechanisms without
being metabolized like endogenous transmitters
 Example: Hydroxyephedrine labeled with carbon-11
HYDROXYEPHEDRINE LABELED WITH
CARBON-11
 One of the most frequently applied PET tracers for
cardiac sympathetic nerve imaging as it shows high
affinity for a common uptake mechanism.
 Can be used for accurate assessment of regional
neuronal defects as it has been shown to distribute
equally within the myocardium in physiologic
conditions.
MORE USES FOR PET IMAGING
Cardiac innervation has also been explored in
heart failure patients who underwent cardiac
transplantation.
PET has also been used to evaluate the
relation between cardiac sympathetic
innervation and ventricular arrhythmias.
FUTURE ROLE OF NUCLEAR IMAGING
In the future s cientists hope to:
 Use nuclear medicine for prevention of overt heart
failure.
 To develop molecular-targeted imaging techniques
that will provide further insight into the
pathophysiology of the failing heart
MICAD
 The Molecular Imaging and Contrast Agent Database
(MICAD) is an online source of scientific information
regarding molecular imaging and contrast agents
(under development, in clinical trials or commercially
available for medical applications) that have in vivo
data (animal or human) published in peer -reviewed
scientific journals.
MICAD
 MICAD is a key component of the “Molecular
Libraries and Imaging” program of the National
Institutes of Health (NIH) Common Fund, designed to
accelerate medical research for disease detection,
diagnosis and therapy. By linking programs in
molecular imaging, molecular probes, and molecular
libraries, the NIH Common Fund provides much
needed support for the development of new, more
specific therapies for a wide range of diseases such
as cancers, Alzheimer’s and Parkinson's diseases.
MICAD
 MICAD is edited by a team of scientific editors and
curators who are based at the National Library of
Medicine, NIH, in Bethesda, Maryland.
 The database includes, but is not limited to, agents
developed for PET, SPECT, MRI, ultrasound,CT, optical
imaging, planar radiography, and planar gamma
imaging. The information on each agent is
summarized in a book chapter format containing
several sections such as Background, Synthesis, in
vitro studies, Animal Studies, Human Studies, and
References.
MICAD
 From http://www.ncbi .nlm.nih.gov/bo oks/NBK5330/ you can download a
CSV file of all of the imaging agents.
 I did this and trimmed the database down to only PET and SPECT agents
that were related to the hear t.